| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 1x8-wasmsimd-splat.c | 65 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_ukernel_1x8__wasmsimd_splat() local
70 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_ukernel_1x8__wasmsimd_splat()
71 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_ukernel_1x8__wasmsimd_splat()
|
| D | 1x8-relu-wasmsimd-splat.c | 65 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_relu_ukernel_1x8__wasmsimd_splat() local
70 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_relu_ukernel_1x8__wasmsimd_splat()
71 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_relu_ukernel_1x8__wasmsimd_splat()
|
| D | 1x8-minmax-wasmsimd-arm-splat.c | 67 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_splat() local
72 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_splat()
73 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_splat()
|
| D | 1x8-minmax-wasmsimd-x86-splat.c | 65 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_splat() local
70 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_splat()
71 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_splat()
|
| D | 3x8-minmax-wasmsimd-arm-splat.c | 99 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
106 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
109 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 97 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
104 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
107 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8-wasmsimd-splat.c | 113 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() local
121 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
125 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neon-dup-ld128.c | 113 const float32x4_t va0c2 = vdupq_lane_f32(vget_high_f32(va0), 0); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() local
117 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c2, vb0123c2); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
121 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c2, vb4567c2); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8-relu-wasmsimd-splat.c | 113 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat() local
121 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
125 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
|
| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 1x8inc-minmax-wasmsimd-arm-splat.c | 69 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_arm_splat() local
74 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_arm_splat()
75 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_arm_splat()
|
| D | 1x8inc-minmax-wasmsimd-x86-splat.c | 67 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_x86_splat() local
72 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_x86_splat()
73 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_x86_splat()
|
| D | 3x8inc-minmax-wasmsimd-arm-splat.c | 101 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat() local
108 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
111 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8inc-minmax-wasmsimd-x86-splat.c | 99 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat() local
106 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
109 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 1x8-wasmsimd-splat.c | 78 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_igemm_ukernel_1x8__wasmsimd_splat() local
83 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_igemm_ukernel_1x8__wasmsimd_splat()
84 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_igemm_ukernel_1x8__wasmsimd_splat()
|
| D | 1x8-relu-wasmsimd-splat.c | 78 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_igemm_relu_ukernel_1x8__wasmsimd_splat() local
83 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_igemm_relu_ukernel_1x8__wasmsimd_splat()
84 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_igemm_relu_ukernel_1x8__wasmsimd_splat()
|
| D | 1x8-minmax-wasmsimd-arm-splat.c | 80 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_arm_splat() local
85 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_arm_splat()
86 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_arm_splat()
|
| D | 1x8-minmax-wasmsimd-x86-splat.c | 78 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_x86_splat() local
83 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_x86_splat()
84 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_x86_splat()
|
| D | 3x8-minmax-wasmsimd-arm-splat.c | 118 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
125 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
128 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 116 const v128_t va0c2 = wasm_v32x4_shuffle(va0, va0, 2, 2, 2, 2); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
123 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c2, vb0123c2)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
126 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c2, vb4567c2)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| /external/XNNPACK/src/f16-gemm/gen/ |
| D | 1x16-minmax-neonfp16arith-ld64.c | 83 const float16x8_t va0c2 = vdupq_lane_f16(va0, 2); in xnn_f16_gemm_minmax_ukernel_1x16__neonfp16arith_ld64() local
85 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c2, vb01234567c2); in xnn_f16_gemm_minmax_ukernel_1x16__neonfp16arith_ld64()
86 vacc0x89ABCDEF = vfmaq_f16(vacc0x89ABCDEF, va0c2, vb89ABCDEFc2); in xnn_f16_gemm_minmax_ukernel_1x16__neonfp16arith_ld64()
|
| D | 1x8-minmax-neonfp16arith-ld64.c | 74 const float16x8_t va0c2 = vdupq_lane_f16(va0, 2); in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64() local
76 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c2, vb01234567c2); in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
|
| /external/XNNPACK/src/f16-gemm/gen-inc/ |
| D | 1x16inc-minmax-neonfp16arith-ld64.c | 85 const float16x8_t va0c2 = vdupq_lane_f16(va0, 2); in xnn_f16_gemminc_minmax_ukernel_1x16__neonfp16arith_ld64() local
87 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c2, vb01234567c2); in xnn_f16_gemminc_minmax_ukernel_1x16__neonfp16arith_ld64()
88 vacc0x89ABCDEF = vfmaq_f16(vacc0x89ABCDEF, va0c2, vb89ABCDEFc2); in xnn_f16_gemminc_minmax_ukernel_1x16__neonfp16arith_ld64()
|
| D | 1x8inc-minmax-neonfp16arith-ld64.c | 76 const float16x8_t va0c2 = vdupq_lane_f16(va0, 2); in xnn_f16_gemminc_minmax_ukernel_1x8__neonfp16arith_ld64() local
78 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c2, vb01234567c2); in xnn_f16_gemminc_minmax_ukernel_1x8__neonfp16arith_ld64()
|
| /external/XNNPACK/src/f16-igemm/gen/ |
| D | 1x16-minmax-neonfp16arith-ld64.c | 94 const float16x8_t va0c2 = vdupq_lane_f16(va0, 2); in xnn_f16_igemm_minmax_ukernel_1x16__neonfp16arith_ld64() local
96 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c2, vb01234567c2); in xnn_f16_igemm_minmax_ukernel_1x16__neonfp16arith_ld64()
97 vacc0x89ABCDEF = vfmaq_f16(vacc0x89ABCDEF, va0c2, vb89ABCDEFc2); in xnn_f16_igemm_minmax_ukernel_1x16__neonfp16arith_ld64()
|
| D | 1x8-minmax-neonfp16arith-ld64.c | 85 const float16x8_t va0c2 = vdupq_lane_f16(va0, 2); in xnn_f16_igemm_minmax_ukernel_1x8__neonfp16arith_ld64() local
87 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c2, vb01234567c2); in xnn_f16_igemm_minmax_ukernel_1x8__neonfp16arith_ld64()
|